An image sensor circuit includes a focal plane array of pixel cells, each one of the cells including either a photogate, photoconductor, or photodiode overlying a charge accumulation region within a substrate for accumulating photo-generated charge. In a conventional four transistor CMOS imager, the active elements of a pixel cell perform: (1) photon to charge conversion; (2) accumulation of image charge; (3) transfer of charge to a floating diffusion node accompanied by charge amplification; (4) resetting the floating diffusion node to a known state before the transfer of charge to it; (5) selection of a pixel for readout; and (6) output and amplification of a signal representing pixel charge. In a three transistor pixel cell the active elements of a pixel cell perform: (1) photon to charge conversion; (2) accumulation of image charge by the photoconversion device; (3) resetting the photoconversion device to a known state before charge accumulation; (4) selection of a pixel for readout; and (5) output and amplification of a signal representing the pixel charge.
Referring to FIGS. 1A and 1B, a semiconductor wafer fragment of a conventional CMOS image sensor four-transistor (4T) pixel 10 is shown. A view of a section of the conventional CMOS image sensor taken along line 1B-1B of FIG. 1A is shown in FIG. 1B. The pixel 10 generally comprises a transfer gate 50 for transferring photoelectric charges generated in a pinned photodiode 21 to a floating diffusion region 25 acting as a sensing node. The floating diffusion region 25 is connected to a reset transistor having a gate 40 for resetting the sensing node. A source follower transistor having a gate 60 is connected to a row select transistor having a gate 80. Impurity doped source/drain regions 22 are provided about gates 40, 60, 80. Spacers 92 may be formed along the sides of gates 40, 50, 60 and 80.
As shown in FIG. 1B, photodiode 21 is illustratively a shallow pinned photodiode just beneath the surface 15 of substrate 20. The pinned photodiode 21 typically has a photosensitive p-n-p junction region comprising a p-type surface region 24 and an n-type photodiode region 26 within a p-type substrate 20. Trench isolation regions 28 are formed in the substrate 20 to isolate the pixels within a pixel array. A translucent or transparent insulating layer 30 may be formed over the pixel 10. Contacts 32 (FIG. 1A) may be formed in the insulating layer 30 to provide an electrical connection to the source/drain regions 22, floating diffusion region 25, and other wiring to connect gate lines and other connections in the pixel 10.
FIG. 2 depicts a conventional four transistor (4T) CMOS image sensor pixel, which optionally employs a polysilicon capacitor (Cpoly) 100. Photons 112 are absorbed just beneath the surface 15 of substrate 20 in the region of the pinned photodiode 21. Electron-hole pairs are generated, and the electrons are collected in n-region 26 of the pinned photodiode 21 as long as the transfer gate 50 is “off”. The pinned photodiode 21 is characterized by a pin potential (Vpin), which is a highest applied voltage of the p-n-p photodiode. Once the transfer gate 50 is activated (i.e., turned “on”), the photon-generated electrons can flow to the floating diffusion region 25 from n-region 26. After a certain period of time, the transfer gate 50 will be switched back to the “off” state.
FIG. 3 is a potential diagram of the conventional 4T CMOS sensor pixel during a light integration period. The full well charge capacity of the pixel is a maximum number of electrons which can be generated and stored in the photodiode 21. Photons 112 transmitted to the photodiode 21 generate electrons in regions 21. The charge capacity of the photodiode 21 to hold the electrons is illustrated by shaded area 120. This area 120 is determined approximately by the pinned potential (VPIN) of photodiode 21 and the photodiode capacity (CPD). When the number of generated electrons reaches the maximum charge capacity, the photodiode 21 is saturated and is unable to further respond to the incident photons 112. The excess charge causes a blooming effect to neighboring pixels as the excess charge moves through substrate 20 into the neighboring pixels. When the electrons from photodiode 21 are transferred to the floating diffusion node 25 by gate 50 turning on, the “capacity” to store electrons is determined by the capacity of the floating diffusion node 25 and that of capacitor 100.
FIG. 4 depicts an exemplary charge transfer operation in the conventional four transistor (4T) CMOS sensor pixel cell. When the fill-well capacity of the photodiode 21 is larger than the charge holding capacity of the floating diffusion node 25, there is charge sharing between the photodiode 21 and floating diffusion node 25. In this case, when the transfer gate 50 goes back to the “off” state, photodiode 21 will retain signal charge which mixes with generated signal charge of the next frame causing image lag. This saturation of the floating diffusion node 25 limits dynamic range of the conventional four transistor (4T) pixel.
As pixels are scaled down in size a lower floating diffusion node capacity is seen which further lowers the dynamic range of a pixel. It is therefore desirable to improve the dynamic range to provide a good output response for low light and also high light signal conditions even when pixels are scaled down. To this end storage capacitors, e.g., capacitor 100, FIG. 2, have been proposed for use with the floating diffusion node to increase charge storage capacity. See, for example, U.S. Pat. Nos. 6,429,470 and 6,204,524 to Rhodes. However, such capacitors while increasing the capacitance of the floating diffusion node, also are relatively inflexible as they are always in circuit with the floating diffusion node. Moreover, additional processing steps are required to form a capacitor coupled to the floating diffusion node.